JOURNAL OF MATERIALS SCIENCE 26 (1991) 3693 3698

High-temperature chlorine corrosion of metals and alloys

A review
Y.-N. CHANG, F.-I. WEI Steel and Aluminium R and D Department, China Steel Corporation, Hsiao Kang, Kaohsiung, Taiwan High-temperature ( > 200~C) corrosion of metals and alloys in atmospheres containing CI2 and/or HCI, which are widely encountered in many industrial environments, is reviewed. Topics include high-temperature corrosion mechanism and kinetics, thermodynamic considerations, and effects of gaseous components in the atmosphere (oxygen, air, water vapour, sulphur dioxide, and nitrogen) on the corrosion.

1. Introduction
Chlorine and hydrogen chloride gases either as reactants, products, by-products or as contaminants are widely encountered in chemical and metallurgical industries such as coal-fired boilers [1], waste incinerators I-2, 3], plastic/polymer decomposition mills [4], vinyl chloride monomer mills [5-9], and recuperators [10]. For example, regardless of the molecular structure of the chloride-containing compounds in the fuel, at the flame temperature it becomes molecular CI> Furthermore, during cooling, C12 may also be converted to HC1 by reacting with either hydrogen or water vapour. Table I is a list of these environments summarized by Elliott et at. [11]. Structural materials in such environments often corrode at accelerated rates at high temperatures because the corrosion products consist of chlorides which generally have lower melting points and boiling points than the oxides of the same metals [12]. In 1976, a review paper on halogen corrosion of metals was published [12] with emphasis on the thermodynamics properties. Since then fundamental studies have complemented work on metals and alloys. Recently, another review on corrosion in chlorine environments was published [13] with emphasis on performance of various materials as well as on materials limitations in such environments. The present paper discusses the high-temperature corrosion of metals and alloys in atmospheres containing C12 and/or HC1. Topics include high-temperature corrosion mechanism and kinetics, thermodynamic considerations, and effects of gaseous components in the atmosphere (oxygen, air, water vapour, sulphur dioxide, and nitrogen) on the corrosion. High temperature here means in excess of 200~ below which the corrosion rates are usually not appreciable.

bolic rate law. In this case, the rate of change of the scale thickness is proportional to the scale thickness itself, i.e.
dx/dt = Kd/X

(1)

where x is the scale thickness, t the time, and Kd is constant controlled by the diffusion process. The parabolic rate law results from the parabola solution of Equation 1. On the other hand, if the metal surface or the phase boundary interface reaction is the ratecontrolling step, then the oxidation rate does not change with time, i.e.
dx/dt = K;

(2)

where K's is a constant controlled by the metal surface or the phase interface reaction, and the oxidation obeys a linear rate law due to the linear line solution of Equation 2. If the diffusion of ions in the scale and the volatiliZation (and/or vaporization) of the scale occur simultaneously during oxidation, then oxidation is the superposition of Equations 1 and 2 except that the constant K's in Equation 2 is negative which means that volatilization results in the reduction of scale thickness, i.e. Equation 2 is modified as
d x / d t = - Ks

(3)

and the rate law equation is

d x / d t = g d / x -- Ks

(4)

2. H i g h - t e m p e r a t u r e corrosion mechanism and kinetics

As discussed previously [14], if the diffusion process in the oxide is the rate-controlling step in high-temperature oxidation, then the oxidation rate obeys a pare0022-2461/91
$03.00 + .12

This is the famous Tedmon equation [15]. However, this type of rate law plays an important role not in high-temperature oxidation, but in hightemperature chlorine corrosion (and even in halogen corrosion) due to the easier volatilization of most metal chlorides than their own" oxides as described above. In this type of corrosion, the scale mentioned above is the chloride of the associated metal. The solution of Equation 4 is
t = - Kd/KZs [ x K s / K a + In (1 - x K s / K d ) ]

(5)

provided the initial condition is that x ----0 at t = 0.

9 1991 Chapman and Hall Ltd.

3693

T A B L E I High-temperature chlorine environments summarized by Elliott et al. [11] High temperature application Temperature range Type of corrosion

(~
Mineral chlorination: precious metals, Ti and Zr production, chlorinators High-temperature chlorination, reactors Vinyl chloride monomer production, ethylene dichloride cracker tubes Production of ethylene dichloride Titanium oxide production, heater tubes in TiCI4 circuit Aluminium melting, Mg removal by C12 injection Fibreglass manufacturing, recuperators Waste incineration, Plant for H z from water Power generation, boilers .. Fluidized bed combustor, heat exchange tubes , 300-900 < 550 < 650 280-480 900 < 850 < 900 < 850 < 900 < 950 850 Clz CI 2 Halide gas HCl, air, ethylene C12 and oxidation Flue gases with CI=, HCI, sulphur, etc. Flue gases including halides and sulphides Flue gases including C12, and sulphur, etc. HC1 Combustion gases including HC1, fuel ash, sulphate/chloride HCI from coal with CaS04, CaO

In the initial period of attack or if the volatilization effect is not significant, the values of x and Ks are small compared with Ko, i~e. xKs/Kd ~ 1, and l n ( 1 - xKJKd) in Equation 5 can be expanded in power series. The result is

..." . ' ' ' ' ' "

9

Parabolic
r a t e law

..-'"
.=_ g ~=

t ~ xZ/2Ka

(6)

which approaches the general second-order parabolic rate law. In the growth period of metal chloride and if the volatilization effect is significant enough, obviously xKs/Kd ~ 1 is no longer valid. Then the rate of change of the chloride thickness obeys Equation 5, i.e. the chloride produced from the ion diffusion volatilizes gradually during the corrosion9 When the thickness approaches the value xf = Ka/K~, Equation 4 becomes dx/dt = 0 and the chloride does not grow further. Therefore, Equation 5 is available only at x < xf. When x exceeds xf, the volatilization at chloride/ atmosphere interface will become the rate-controlling step and the linear rate law of corrosion, Equation 3, is satisfied. Thus, the total weight of the metal decreases gradually. From the discussiOn above, one can expect that the typical weight change of the metal with time is like curve A in Fig. i, where xf (or tf), which was defined above the critical thickness (or time) when the scale no longer grows, depends on the temperature, atmosphere, and the chloride characteristics. The temperature always promotes the volatilization effect. If the growth of chloride is not fast enough to supply the amount required for volatilization, the corrosion rate law will obey Equation 3 (curve B, Fig. 1). For exmple, the corrosion of iron in 1.46 vol% HC1/N2 (1 atm) at temperatures below 500~ obeyed a parabolic rate law due to the formation of a protective FeC12 film on the metal surface [16]. As the temperature increased to 550 ~ the rate law was like curve A in Fig. 1. As it increased further to 800 ~ the rate law became like curve B. Similar kinetics also resulted for the carrosion of iron in 1 vol % C12/N2 (1 atm) except that the transition temperatures were different. Other investigations [5-9, t l, 17-281, using thermogravimetry techniques, also obtained Corrosion kinetics typical of the curves of Fig. 1. Usually the diffusion of ions in chloride is fast enough only at temperatures above 0.5Tr,, where Tm is the melting point of the chloride. Furthermore, volat3694

Time

,i
_o

Figure 1 Typical curves of weight change of metal with time in

high-temperature chlorine corrosion.

T A B L E II Common binary metal chloride melting points (T=) and temperatures (T4) at which the chloride vapour pressure is 10 -4 atm Chloride Tm (~ 303 676 1030 740 1150 820 434 678 T4 (~ 167 536 607 587 61~ 741 132 474 Reference [12] [12] [12] [12] [12] [12] [12] [12]

Feel 3 FeCI2 NiCI2 CoC12 CrC13 CrCI2 HfCI4 PdC12 ThCI4 CeC13

770 800

564 911

[29] [29]

ilization is significant only at vapour pressures above 10 -4 atm. Table II lists the T= and T4 (the temperature at which the vapour pressure is 10 -4 atm) values of some common metal chlorides summarized previously [12, 29]. However, we believe that T4 plays a

more important role than Tm in high-temperature chlorine corrosion.

3. T h e r m o d y n a m i c c o n s i d e r a t i o n s In high-temperature oxidation, the temperature dependence of the free energies of formation of oxides (Ellingham diagram) play an important role because one can predict the stable oxides at a particular temperature. Similarly, in high-temperature chlorine corrosion, stable chlorides can also be predicted from their free energies of formation as discussed earlier [12, 13]. However, in a mixture of more than one gas component, the situation is more complex. For example, considering pure iron in the mixture of hydrogen chloride and oxygen, the possible reactions are as follows.
nFe + m/2
O 2 =

chlorination leads typically to the formation of FeC13 by Reaction 11 as well as the formation of FeC12 by Reaction 10, as demonstrated by Ihara et al. [17]. Similarly, the possible reactions of chromium in the mixture of HC1 and 02 are as follows. 2Cr + 3/2 02 = Cr203 Cr + 2HC1 = CrC12 + H2 CrC12 + HC1 = CrC13 + 1/2 H2 Cr + 2HC1 + 1/2 O2 = CrC12 + H 2 0 CrC12 + HC1 + 1/4 02 = CrC13 + 1/2 H 2 0 1/3 Cr203 + 2HC1 = 2/3 CrCI3 + H20 CrC12 + 3/4 02 = 1/2 Cr203 + CIz 2/3 CrC13 + 1/2 02 = 1/3 Cr203 + C12 (15) (16) (17) (18) (19) (20) (21) (22)

Fe,O,,

(7) (8) (9) (10) (11) (12) (13) (14)

Fe + 2HC1 = FeC12 + H2 FeCI2 + HC1 = FeCI3 + 1/2 H2 Fe + 2HCI + 1/2 O2 = FeC12 + H20 FeCI2 + HC1 + 1/4 02 = FeC13 + 1/2 H 2 0 1/3 Fe203 + 2HC1 = 2/3 FeC13 + H 2 0 FeCI2 + 3/4 O2 = 1/2 Fe203 + CI2 2/3 FeC13 + 1 / 2 0 z = 1/3 FezO3 + C12

Green sublimate corrosion product CrCI3-6H20 after the corrosion of chromium in HC1/20-75 vol % 02 mixture at 550-800 ~ observed [9] might come from the oxy-chlorination Reaction 19. Oxy-chlorination phenomena were also observed in other investigations [-4, 5, 7, 8, 11, 18-23, 30-33].

4. Effects of gaseous c o m p o n e n t s on t h e

corrosion
Because many industrial atmospheres may also contain other flue gases in addition to C12 a n d / o r HC1, it is necessary to visualize the effects of these gaseous components on the chlorine corrosion. The role of chlorine itself in flue gas corrosion has been reviewed recently [13].

Fig. 2 shows the temperature dependence of the standard free energies of Reactions 7-14 calculated from the thermodynamic data [29]. From the diagram one can expect that, in pure HC1, FeC12 is formed by Reaction 8, but further chlorination (Reaction 9) is thermodynamically impossible. In contrast, when the reaction gases contain both HC1 and 02, the oxy-

20
i E

I0

Iz
o -I0

-~
v

~=
o

-20 ,s -30 -40 & -50

IZl

11

-60 -70
I

500 400

500 600 700 800 Temperature (K)

900 1000

Figure 2 Changesin standard free energies of Reactions 8-14 with

temperature calculated from [29], in which sublimation, melting and vaporization of reactants and products are not taken into account.

4.1. O x y g e n a n d air As described above, the reaction kinetics of iron in 1.46vo1% HC1/N2 (1 atm) below 500~ obeyed a parabolic rate law due to the protective FeCI2 film. Ihara et al. [17] also found that in pure HC1 gas at 300-800 ~ the corrosion behaviour was determined by the formation and sublimation of FeC12. Addition of 20-50 vol % 02 to HCI greatly accelerated the corrosion because of the formation and sublimation of a low melting point volatile FeC13 via oxy-chlorination such as Reaction 11 described above. Similar results were also observed by the same investigators [7] on Fe-Ni alloys in the same gas mixture at temperatures greater than 500 ~ and by Jacobson [18] on reaction of iron with 1 vol % HC1/0-50 vol % O2/Ar mixture at 550~ Above 50 vol % 02, the accelerating effect became less pronounced and it could even be reversed as shown in Fig. 3 [17] due to the formation of surface oxide films. In C12, however, oxygen had little effect on the chlorination rate of iron below about 300~ although it prevented vaporization of FeC13 above this temperature due to the competitive formation of an iron oxide film too [34, 35]. In HC1 gas the corrosion rate of chromium at 400-800 ~ is determined by the formation and vaporization of CrC12 scale to some extent up to 600 ~
3695

i0 A

102

10 3

<
tt%J

800o C

I01

PE

=n
0 L c 0

102"

'E

9
100

g
o 101

2 g
0 C~

\oOOO
i0 -I 10 -2

[]
1o 0

/
O
I
| i

300~
O

i0 -I

0
, I I~ I I

20

50

75

I00

20 50 75 0 2 content (vo[ %)

100

0 2 content (voL %)
Figure 3 Change in average corrosion rates of iron in HC1, O2 and
their mixtures as a function of O2 content, reproduced from [-17]. The average corrosion rate was estimated from the weight loss after corrosion products were removed from specimens which were exposed to reaction gases for 5 h.

Figure 4 Changes in average corrosion rates of chromium in flowing HC1, O2 and their mixtures as a function of O2 content, reproduced from [9]. The average corrosion rate was estimated from specimens which were exposed to reaction gases for 5 h.

Addition of 20-75 vo1% 0 2 led to suppression of corrosion up to about 600 ~ due to the formation of Cr203, but to catastrophic corrosion at higher temperatures as shown in Fig. 4 [9] due to the vaporization of CrC13 and water vapor which might also come from oxy-chlorination. In mixtures of C12 and O2, the corrosion of chromium was inhibited even at temperatures as high as 700 ~ through the formation of an outer Cr203 layer and of CrOzClz as the only volatile product [19]. The corrosion rates of Fe-Cr alloys in HC1 at 500 650 ~ are determined by the formation and sublimation of both FeCI2 and CrC12. Because of the formation of a protective corundum-type oxide scale ( a - [Fe(Cr)]203) on alloys with chromium content ~> 13wt%, addition of 20-50 vol% O2 effectively suppressed the corrosion at relatively low temperatures [8]. This type of corundum-type oxide scale was also observed on austenitic stainless steels at 450-500 ~ inthe same range of gas mixtures [5]. The temperature range for the protective scale formation is raised with increasing chromium content in the alloy and oxygen content in the gas mixture. Further rise in temperature accelerates corrosion due to sublimation of FeC13 and CrC13 as a result of oxy-chlorination. For example, the oxide scale [-Fe(Cr)]203 formed on Fe-13 wt% Cr alloy above 500~ in 80 vol% HC1 + 20 vol % Oz was depleted of chromium due to the sublimation of CrC13, and became less protective 3696

[8]. No data on corrosion of Fe-Cr alloys in gas mixture of Clz and Oz have been published. The oxy-chlorination products of molybdenum are very volatile. Therefore, the corrosion rate of SUS 316 steel in gas mixtures of HC1 and 20-50 vol % O2 at temperatures lower than 420~ was reported to be significantly high [-5] because the protective scale became defective by sublimation of the oxy-chlorides of molybdenum. At still higher temperatures the detrimental effect of molybdenum in the steel was not significant because all the alloy constituents also suffered oxy-chlorination in the high-temperature range. Because of exclusive formation of NiClz scale at 400-700 ~ the presence of oxygen in HCl-containing atmospheres, regardless of the gas compositions, appears to have a negligible effect on corrosion of nickel as demonstrated by Ihara et al. [6]. But its attack at 927 ~ was predicted by McNallan and Liang [-36] to decrease with increasing oxygen partial pressure in chlorine environments. Corrosion in gas mixture of CI2 and O2 at lower temperature, as in the case of pure HC1, proceeds mainly via formation and volatilization of NiClz. However, at much higher temperatures the rate of corrosion of nickel in CI 2 is considerably accelerated by the presence of oxygen probably due to the following two reactions 2Ni(s) + C l 2 ( g ) = NiCl,as 2 NiClads + 1/2 O2(g) = 2 NiO(s) + NiClz(g) (23) (24)

150
"7
t~j

T:

800 ~ C

'~E
b = o g
o c.l

100

~
,o""
~/'~
I I

0 ......

0--

T= 750 ~ C

5O

from the atmosphere through the outer NiO layer to react with inner AI and Cr.. The reaction product H2 resided at the alloy/oxide (A1203, Cr203) interface and induced oxide cracking when it attained a certain partial pressure. Spalling of the oxide corrosion products was also observed during the corrosion of ]ncoloy 800 (32Ni-21Cr 46Fe) in moist air containing HC1 at 800 ~ [31]. But no direct evidence can demonstrate that it came from H 2 0 or from HC1. 4.3. S u l p h u r d i o x i d e a n d n i t r o g e n Woodford and Bricknell [44] found that penetration and embrittlement of grain boundaries by sulphur occurred in nickel and nickel-base superalloy. It affected not only the corrosion of flue gas at 900 ~ but also the mechanical propertieS of the materials. It was also observed [1] that 0.18 vol % SOz in HCIcontaining flue gas promoted susceptibility to cracking at 540 ~ of ferritic steels by formation of FeS at the metal/scale interface and of austenitic steels by sulphur segregation at grain boundaries. However, Prescott a n d co-workers [33, 45] reported that 0.86 vol % SO2 in HCl-containing flue gas did not affect the corrosion of metals and alloys at 900 ~ because the partial pressure of sulphur in that environment was too low to form sulphide at that temperature. In addition, 0.5 vol % SOz in a combustion HCl-containing flue gas probably did not affect the corrosion of mild and low alloy steels at 4002500 ~ although the metallographic examination revealed FeS in the outer parts of the scale [46]. The effect of other flue gas such as nitrogen is dilution of chlorine only and the corrosion is thus reduced as reported much earlier [35].

Pc[2 34 t0rr v= 15cm rain-1

I

~S

Oq

200

400 P02 (torr)

600

800

Figure 5 Effect of 02 pressure on corrosion of Ni in Ct 2 at 750 and

800JC [37]. I t 0 r r - 1.333 x 102 Pa.

The mechanism described by these reactions involves formation and oxidation of NiC1 which was detected spectroscopically to be formed as an intermediate at high temperatures [37]. The accelerating effect of oxygen on chlorination of nickel at higher temperatures is illustrated by Fig. 5 reproduced from [37]. In summary, the effect of oxygen is to produce an oxide scale or oxy-chloride scale. Oxides are beneficial to the corrosion resistance. However, the effect of oxy-chloride is detrimental if it is more volatile than the scale produced from the atmosphere without oxygen, while it is beneficial if it is less volatile. The effect of air on corrosion seems similar to that of oxygen. An additional observation is that found by Elliott et al. [21, 23] on aluminium-containing alloy in air/2 vol % C12 mixture at 900 and 1000~ Alumina (A1203) is preferentially formed by air and the formation of volatile A1C13 is inhibited. The corrosion rate of the alloy is thus reduced. However, we expect that this phenomenon would also occur in mixtures of oxygen and chlorine.

5. Conclusions
1. High-temperature corrosion consists mainly of the diffusion of ions in the scale and the volatilization of the scale. The rate-controlling step from the corrosion kiv~tics which results from the Tedmon equation can be determined. 2. In a mixture of more than one gas component, free energies of formation of all possible reactions should be considered for prediction of reaction products from the thermodynamic viewpoint. Oxychlorination often occurs in those chlorine atmospheres containing oxygen. 3. Effect of oxygen and/or air is to produce an oxide scale or oxy-chloride scale. Oxides are beneficiM to the corrosion resistance. However, the effect of oxychloride is detrimental if it is more volatile than the scale produced from the atmosphere without oxygen, while it is beneficial if it is less volatile. In some cases alumina is preferentially formed by oxygen and the formation of volatile A1C13 is in.hibited. 4. Effect of water vapour on the corrosion resistance of most metals is beneficial due to the formation of a protective oxide. However, in some cases such as Ni-Cr-A1 alloy in HCl-containing moist air [42, 43], oxide cracking can occur due to the hydrogen produced from the reaction of H / O with A1 and Cr at the alloy/oxide interface,
3697

4.2. W a t e r v a p o u r The effect of water vapour on the chlorine corrosion was examined early in 1956 [34, 38], and was found that the corrosion rates of most tested metals, such as iron and its alloys, stainless steels, aluminium, and tantalum, were greatly diminished in the presence of water vapour because of the protective oxide film up to a certain limiting temperature. Recently, Gesmundo et al. [39] also reported that, during the corrosion of ternary Ni 29Mo-4Ee alloy, water vapour in an HCl-containing atmosphere promoted the formation of NiO and inhibited the formation of NiC12. A similar phenomenon was also expected during the corrosion of Ni-containing alloy Hastelloy B [40]. All these phenomena might be attributed to the fact that oxides of most metals are more stable than their own chlorides [12, 41]. However, during the corrosion of Ni-Cr-A1 alloy in HCl-containing moist air [42, 43], NiO was permeable to water vapour and impermeable to hydrogen. Therefore, H 2 0 penetrated

5. In some cases, sulphur dioxide is detrimental to the chlorine corrosion resistance because of the susceptibility to embrittlement or cracking, while in other cases it does not affect the corrosion. Effect of nitrogen is dilution of chlorine only and the corrosion rate is thus reduced.

24.

25~ 26.

References
1. D. B. MEADOWCROFT and M. I. MANNING, in "Corrosion Resistant Materials for Coal Conversion Systems" (Applied Science, London, New York, 1983) pp. 87, 105. 2. H . H . KRAUSE, in "High Temperature Corrosion in Energy Systems", Detroit, MI, 1984, edited by M. F. Rothman (The Metallurgical Society of AIME, Warrendale, PA, 1985) p. 83. 3. W.C. FORT and L. W. R. DICKS, in Proceedings of a Conference on Corrosion '85 (NACE, Boston, MA, 1985) Paper 12. 4. G. MARSH and P. ELLIOTT, High Temp. Teehnol. 1 (1982) 115. 5. Y. IHARA, K. SAKIYAMA and K. HASHIMOTO, in "Proceedings of JIMIS-3, High Temperature Corrosion Transactions", Suppl. Jpn. Inst. Metals 24 (1983) 669. 6. Y. IHARA, H. OHGAME, K. SAKIYAMA and K. HASHIMOTO, Corrosion Sci. 22 (1982) 901. 7. ldem, Trans. Jpn. Inst. Metals 23 (1982) 682. 8. Y. IHARA, H. OHGAME and K. HASHIMOTO, ibid. 25 (1984) 96. 9. Y. IHARA, H. OHGAME, K. SAKIYAMA and K. HASHIMOTO, Corrosion Sci. 23 (1983) 167. 10. G. Y. LAI, M. F. ROTHMAN, S. BARANOW and R. B. HERCHENROEDER, J. Metals 35 (1983) 24. 11. P. ELLIOTT, A. A. ANSARI and R. NA13OVI, in "High Temperature Corrosion in Energy Systems", Detroit, MI, 1984, edited by M. F. Rothman (The Metallurgical Society of AIME, Warrendale, PA, 1985) p. 437. 12. M.G. FONTANA and R. W. STAEHLE (eds), "Advances in Corrosion Science and Technology", Vol. 5 (Plenum Press, New York; 1976) p. 55. 13. z . A . FOROULIS, Anti-Corrosion 35 (1988) 4. 14. Y.N. CHANG and F. I. WEI, J. Mater. Sci. 24 (1989) 14. 15. C.S. TEDMON Jr, J. Electrochem. Soc. 113 (1966) 766. 16. Z.A. FOROULIS, in "Proceedings of JIMIS-3, High Temperature Corrosion Transactions", Suppl. to Jpn. Inst. Metals 24 (1983) 699. 17, Y. IHARA, H. OHGAME, K. SAKIYAMA and K. HASHIMOTO,.Corrosion Sci. 21 (1981) 805. 18. NI S. JACOBSON, Oxid. Metals 26 (1986) 157. 19. K. REINHOLD and K. HAUFFE, J. Electrochem. Soc. 124 (1977) 875. 20. S. BARANOW, G. Y. LAI, M. F. ROTHMAN, J. M. OH, M. J. McNALLAN and M. H. RHEE, in Proceedings of a Conference on Corrosion '84 (NACE, New Orleans, LA, April 1984) Paper 16. 21. P. ELLIOTT, A. A. ANSARI, R. PRESCOTT and M. F. R O T H M A N, in Proceedings of a Conference on Corrosion '85 (NACE, Boston, MA, 1985) Paper 13. 22. J.M. OH, M. J. McNALLAN, G. Y. LAI and M. F. ROTHMAN, Metall. Trans. 17A (1986) 1087. 23. P. ELLIOTT, A. A. ANSARI, R. PRESCOTT and 27. 28. 29.

30.

31. 32. 33. 34. 35. 36.

37. 38. 39. 40. 41. 42. 43.

44. 45. 46.

M. F. ROTHMAN, Corrosion NACE 44 (1988) 544. W.W. LIANG and K. H. YUN, in "Corrosion-Erosion Behaviour of Metals", St Louis, Missouri, 1978, edited by K. Natesan (The Metallurgical Society of AIME, Warrendale, PA, 1978) p. 61. R. P. VISWANATH, D. REIN and K. KAUFFE, Werk. Korros. 31 (1980) 778. M.H. RHEE, M. J. McNALLAN and M. F. ROTHMAN, in "High Temperature Corrosion in Energy Systems", Detroit, MI, 1984, edited by M. F. Rothman (the Metallurgical Society of AIME, Warrendale, PA, 1985) p. 483. M.J. MALONEY and M. J. McNALLAN, Metall. Trans. 16B (1985) 751. N.S. JACOBSON, M. J. McNALLAN and Y. Y. LEE, ibid. 17A (1986) 1223. O. KUBASCHEWSKI and C. 13. ALCOCK, in '~Metallurgical Thermochemistry", 5th Edn, Vol. 24 (Pergamon Press, Oxford, 1979) p. 358 P. ELLIOTT, A. A. ANSARI and G. MARSH, in "International Congress on Metallic Corrosion", Vol. 3 (National Research Council of Canada, Toronto, Canada, June 1984) p. 330. P. ELLIOTT and G. MARSH, Corrosion Sci. 24 (1984) 397. F.H. SCOTT, R. PRESCOTT, P. ELLIOTT and M. H. J. H. AL'AT1A, High Temperature Technology 6 (1988) 115. R. PRESCOTT, F. H. STOTT and P. ELLIOTT, Oxid. Metals 31 (1989) 145.K. L. TSEITLIN and V. A. STRUNKIN, J. Appl. Chem. USSR 29 (1956) 1793. K.L. T S E I T L I N a n d v . A. STRUNKIN, ibid. 31(1958) 1832. M.J. McNALLAN and W. W. LIANG, in "High Temperature Corrosion in Energy Systems", Detroit, MI, 1984, edited by M. F. Rothman (The Metallurgical Society of AIME, Warrendale, PA, 1985) p. 457. K. HAUFFE and J. HINRICHS, in "Oxidation of Metals and Alloys" (ASM, Metals Park, OH, 1971) p. 87. K . L . TSEITLIN, J. Appl. Chem. USSR 29 (1956) 1281. F. GESMUNDO, C. De ASMUNDIS, F. C O E N - P O R I S I N I and E. RUEDL, Br. Corros. J. 15 (1980) 14. E. RUEDL, F. COEN-PORISINI, T. SASAKI and C. De ASMUNDIS, J. Nucl. Mater. 101 (1981) 135. P. ELLIOTT, C. J. TYREMAN and R. PRESCOTT, J. Metals 37(7) (1985) 20. M. K. HOSSAIN, A. C. NOKE and S. R. J. SAUNDERS, Oxid. Metals 12 (1978) 451i D. COUTSOURADIS, P. FELIX, H. FISCHMEISTER, L. HABRAKEN, Y. LINDBLOM and M. O. SPEIDEL, in "High Temperature Alloys for Gas Turbines" (Applied Science, London, 1978) p. 239. D.A. W O O D F O R D and R. H. BRICKNELL, Seripta Metall. 17 (1983) 1341. F. H. STOTT, R. PRESCOTT and P. ELLIOTT, Werk. Korros. 39 (1988) 401. S. 13ROOKSandD. 13. MEADOWCROFT, in "High Temperature Corrosion in Energy Systems", Detroit, MI, 1984, edited by M. F. Rothman (The Metallurgical Society of AIME, Warrendale, PA, 1985) p. 515.

Received 30 April and accepted 29 October 1990

3698